fix: inherit agent model in Skill Forge LLM calls

raven/config/raven.py

@@ -964,8 +964,8 @@ class SkillForgeConfig(_Base):
     with RRF output size (local_pool_top_k + mass_pool_top_k dedupe)."""
 
     llm_gate_model: str | None = None
-    """Optional model override for gate calls. ``None`` → use the
-    provider's default chat model (typically the agent's main model)."""
+    """Optional model override for gate calls. ``None`` uses the active
+    agent model."""
 
     llm_gate_temperature: float = 0.0
     """Sampling temperature for gate calls. 0.0 for deterministic

raven/context_engine/factory.py

@@ -108,6 +108,7 @@ def build_context_engine(
 
     rewriter, gate = _build_rewriter_and_gate(
         provider=provider,
+        model=model,
         skill_forge_config=skill_forge_config,
         skill_forge_router_config=skill_forge_router_config,
     )
@@ -216,6 +217,7 @@ def _build_router(
 def _build_rewriter_and_gate(
     *,
     provider: LLMProvider,
+    model: str,
     skill_forge_config: "SkillForgeConfig | None",
     skill_forge_router_config: "SkillForgeRouterConfig",
 ) -> "tuple[QueryRewriter | None, LLMGateFilter | None]":
@@ -236,6 +238,7 @@ def _build_rewriter_and_gate(
     if bool(getattr(skill_forge_config, "rewrite_enabled", False)):
         rewriter = QueryRewriter(
             provider,
+            model=model,
             max_tokens=int(getattr(skill_forge_config, "rewrite_max_tokens", 8192) or 8192),
         )
 
@@ -248,7 +251,7 @@ def _build_rewriter_and_gate(
             provider,
             max_select=int(getattr(skill_forge_config, "llm_gate_max_select", 2) or 2),
             legacy_top_k=int(skill_forge_router_config.top_k or 5),
-            model=getattr(skill_forge_config, "llm_gate_model", None) or None,
+            model=getattr(skill_forge_config, "llm_gate_model", None) or model,
             temperature=float(getattr(skill_forge_config, "llm_gate_temperature", 0.0)),
             max_tokens=int(getattr(skill_forge_config, "llm_gate_max_tokens", 8192) or 8192),
         )

raven/memory_engine/skill_forge/rewriter.py

@@ -67,10 +67,12 @@ def __init__(
         self,
         provider: "LLMProvider",
         *,
+        model: str | None = None,
         max_tokens: int = 8192,
         temperature: float = 0.3,
     ) -> None:
         self._provider = provider
+        self._model = model
         self._max_tokens = max_tokens
         self._temperature = temperature
 
@@ -85,6 +87,7 @@ async def analyze(self, query: str) -> RewriteResult:
             resp = await asyncio.wait_for(
                 self._provider.chat_with_retry(
                     messages=[{"role": "user", "content": prompt}],
+                    model=self._model or None,
                     max_tokens=self._max_tokens,
                     temperature=self._temperature,
                 ),

tests/test_phase_a_default_engine.py

@@ -25,6 +25,7 @@
     ContextConfig,
     HubSourceConfig,
     MemoryConfig,
+    SkillForgeConfig,
     SkillForgeRouterConfig,
 )
 from raven.context_engine import ContextAssembler
@@ -81,18 +82,21 @@ def _build_engine(
     backend=None,
     hub_endpoint: str | None = None,
     memory_config: MemoryConfig | None = None,
+    model: str = "stub",
+    skill_forge_config: SkillForgeConfig | None = None,
 ) -> ContextAssembler:
     builder = ContextBuilder(workspace=tmp_path)
     engine = build_context_engine(
         workspace=tmp_path,
         config=ContextConfig(),
         builder=builder,
         provider=_StubProvider(),
-        model="stub",
+        model=model,
         context_window_tokens=8192,
         get_tool_definitions=_stub_get_defs,
         backend=backend,
         memory_config=memory_config or MemoryConfig(),
+        skill_forge_config=skill_forge_config,
         skill_forge_router_config=SkillForgeRouterConfig(
             hub=HubSourceConfig(endpoint=hub_endpoint),
         ),
@@ -102,10 +106,14 @@ def _build_engine(
 
 
 def _router_sources(engine: ContextAssembler):
-    skills = next(b for b in engine._builders if isinstance(b, SkillsSegmentBuilder))
+    skills = _skills_builder(engine)
     return [type(s) for s in skills._router._sources], skills._router._sources
 
 
+def _skills_builder(engine: ContextAssembler) -> SkillsSegmentBuilder:
+    return next(b for b in engine._builders if isinstance(b, SkillsSegmentBuilder))
+
+
 def _memory_builder(engine: ContextAssembler) -> MemorySegmentBuilder:
     return next(b for b in engine._builders if isinstance(b, MemorySegmentBuilder))
 
@@ -124,6 +132,26 @@ def test_returns_assembler_without_backend(self, tmp_path: Path) -> None:
         assert isinstance(engine, ContextAssembler)
         assert _memory_builder(engine)._backend is None
 
+    def test_skill_forge_llms_inherit_agent_model(self, tmp_path: Path) -> None:
+        engine = _build_engine(
+            tmp_path,
+            model="minimax/minimax-m3",
+            skill_forge_config=SkillForgeConfig(),
+        )
+        skills = _skills_builder(engine)
+        assert skills._rewriter._model == "minimax/minimax-m3"
+        assert skills._gate._model == "minimax/minimax-m3"
+
+    def test_gate_model_override_takes_priority(self, tmp_path: Path) -> None:
+        engine = _build_engine(
+            tmp_path,
+            model="minimax/minimax-m3",
+            skill_forge_config=SkillForgeConfig(llm_gate_model="anthropic/claude-sonnet-4-5"),
+        )
+        skills = _skills_builder(engine)
+        assert skills._rewriter._model == "minimax/minimax-m3"
+        assert skills._gate._model == "anthropic/claude-sonnet-4-5"
+
 
 # ---------------------------------------------------------------------------
 # SkillForgeRouter assembly — which sources are present

tests/test_skill_forge_rewriter.py

@@ -64,6 +64,13 @@ async def test_analyze_returns_rewritten_query() -> None:
     assert result.rewritten_query == "generate pdf reports"
 
 
+async def test_analyze_forwards_configured_model() -> None:
+    provider = _StubProvider(json.dumps({"need_retrieval": False}))
+    rewriter = QueryRewriter(provider, model="minimax/minimax-m3")
+    await rewriter.analyze("hello there")
+    assert provider.calls[0]["model"] == "minimax/minimax-m3"
+
+
 async def test_analyze_handles_code_fence_wrapping() -> None:
     provider = _StubProvider('```json\n{"need_retrieval": true, "rewritten_query": "trim"}\n```')
     result = await QueryRewriter(provider).analyze("verbose query")